Menu
Sister Sites

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

How do you deal with your losses? 87 replies

Do you believe tight stop-losses cause losses in the long run? 76 replies

Do you believe fixed stop-losses cause losses in the long run? 16 replies

Deal or no Deal Syndrome and other conditions that affect traders 6 replies

Are you a small speculator? Then I know where your Stop Losses are! :) 11 replies

How to deal with the losses??

Thanks, Ubeee, I always enjoy your posts. Sounds like you're way ahead of where I'm at currently, in terms of trading profitably (with a good measure of
consistency).

For me the most difficult part is not mindset, but simply finding non-random patterns in price movement, that recur often enough to provide a decent edge. There are days when I feel I'm making great progress in this respect, and days when I feel I'm no further ahead than when I first started. My biggest obstacle is that I lack accurate test data, hence all of my back-testing is performed manually. Ironic perhaps, considering that writing computer software has been my career since I left high school in 1979.

Best wishes,
David
id like to be intelligente to make my own neural network EA, but the best i did was overoptimize the systems that came with Neuroshell, Trading Solutions and one another that i don't remember the name now.

man, i wasted a lot of time with it
Quoting giraia_br
id like to be intelligente to make my own neural network EA, but the best i did was overoptimize the systems that came with Neuroshell, Trading Solutions and one another that i don't remember the name now.

man, i wasted a lot of time with it

giraia_br,
there are tons of ways to avoid curve fitting problems both specific to neural networks or classical systems.
I dont't want to drive this thread off topic, but just to make one single example try to cut all of your neural network wins by a factor of 5 but leave the losses as they are. If you still manage to have a neural network that is profitable, believe me, it will have better performance both on out of sample data and on real trading (when of course you do not cut the wins!) than on training data.
Just try. You will be very happily surprised.
.Ubeee.
Quoting ubeee
You are not well informed about this, believe me. Just google it and you will discover a universe.
Just to give you a good start try this:
http://en.wikipedia.org/wiki/James_Harris_Simons
He is strongly in profits (actually his hedge fund is universally considered the most successful hedge fund of all time). And he has been in big profits for decades, not just months or years. Oh, and by the way, he hires only mathematicians and physicists, never economists or traders.

Sorry, but I prefer to use my brain for what it's best at: creativity. As I see it automation is for automaton, not for human brains. It would be really a pity to use the most powerful tool we know in the entire universe (at least so far) to let it do just some automatic things!

Yes, absolutely! In fact I strongly think that it is a totally false myth believing that whoever automate her/his trading does it as a workaround for her/his lack of discipline, or any other quality needed to be a good trader. Also I like to point out again that while discipline is surely something extremely important, and very necessary, in trading, it is by no mean anything special in trading if compared to any other work that imply risk.
You need discipline to be a pilot, to be a fireman, to be a doctor, a policeman and so on. But again, discipline alone will never make you a pilot, not a fireman, neither a doctor, or a policeman, and surely discipline alone will never make you a trader, let alone a profitable one.
So if I let my systems trade automatically in my place it is not to avoid any psych problems I have, or to build a surrogate of the discipline needed to trade, it is just to put myself in what I feel are the optimal conditions to get the maximum profits I could from the markets, while having the maximum free time to enjoy my life as I can. That's all.
Do you really think that using our brain this way is not that good?
It is true that I am not especially well informed about the most recent developments in these models and I had not heard of this Harris Simons character. The guy appears to be a genius with a Nobel prize winning network of connections and stacks of capital to back it up - and that is what it takes, yet even that isn't enough. LTCM fit into the same category and the flaws of these models (designed by a gang of extremely intelligent and well remunerated individuals recruited from the top educational institutions around the world) surfaced when the fund imploded due to some error which they put down to a '10-sigma event'. Many hedge funds who run these models are struggling badly in the current market, citing the same excuses.

Anyway, my point is that if those with seemingly unlimited resources have a hard time of it then it seems a little unlikely that someone without those resources is going to fare any better. Perhaps I underestimate your intelligence, but I suspect many of these amateur EA's are based on MA/indicator crosses etc which is a different debate altogether. I certainly wish you no ill will though and hope you prove me wrong.
Quoting birdt
...The guy appears to be a genius with a Nobel prize winning network of connections and stacks of capital to back it up - and that is what it takes, yet even that isn't enough.
birdt,

my intention was not to say that automatic trading is superior in any way (more profitable or easier to master) than discretionary trading. I think that we simply cannot be sure neither of this or the contrary.
I just wanted to point out that it is not true (and this is for sure) that
"Banks and the serious big hitters have pumped billions and hired some of the greatest mathematical minds to develop models that outperform the markets and ... they have not succeeded."
In fact as you can read here:
http://en.wikipedia.org/wiki/Algorithmic_trading

"A third of all EU and US stock trades in 2006 were driven by automatic programs, or algorithms, according to Boston-based consulting firm Aite Group LLC. By 2010, that figure will reach 50 percent, according to Aite.[2]

In 2006 at the London Stock Exchange, over 40% of all orders were entered by algo traders, with 60% predicted for 2007. American markets and equity markets generally have a higher proportion of algo trades than other markets, and estimates for 2008 range as high as an 80% proportion in some markets. Foreign exchange markets also have active algo trading (about 25% of orders in 2006).[3] Futures and options markets are considered to be fairly easily integrated into algorithmic trading[4], with about 20% of options volume expected to be computer generated by 2010.[5] Bond markets are moving toward more access to algorithmic traders.[6]"

So not only this field is alive, it is also rapidly increasing. And if it is increasing maybe there could be something valuable in this approach.
Well, now I don't know if automatic trading requires more intelligence or Nobel prizes or money or resources in general than discretionary trading.
What I know for sure is only these:
1) At my age you know what are your strengths and weaknesses and I know for sure that I'm absolutely not more intelligent than the average person on this planet.
2) In years of testing, trying + live, real trading I've never been able to master any discretionary system such in a way to be profitable for more than 2 consecutive weeks.
3) But at the same time I've developed 10 automatic systems that show consistent profitability with small drawdowns over 8 years of training data, 6 months of out of sample data and (for three of them, the only three that I decided to trade live) 7 months of profitable live trading.
You know, all of this does really prove anything at all. Too few data and too short time to prove anything. But, please, tell me sincerely, if you was in me, given those facts, would you go for discretionary trading or automatic?
.Ubeee.
Quoting ubeee
"A third of all EU and US stock trades in 2006 were driven by automatic programs, or algorithms, according to Boston-based consulting firm Aite Group LLC. By 2010, that figure will reach 50 percent, according to Aite.[2]

In 2006 at the London Stock Exchange, over 40% of all orders were entered by algo traders, with 60% predicted for 2007. American markets and equity markets generally have a higher proportion of algo trades than other markets, and estimates for 2008 range as high as an 80% proportion in some markets. Foreign exchange markets also have active algo trading (about 25% of orders in 2006).[3] Futures and options markets are considered to be fairly easily integrated into algorithmic trading[4], with about 20% of options volume expected to be computer generated by 2010.[5] Bond markets are moving toward more access to algorithmic traders.[6]"
I wonder how many of those fall into the 95% that are unsuccessfull at trading ;-)
Visit my journal here.
Quoting ubeee
birdt,

my intention was not to say that automatic trading is superior in any way (more profitable or easier to master) than discretionary trading. I think that we simply cannot be sure neither of this or the contrary.
I just wanted to point out that it is not true (and this is for sure) that
"Banks and the serious big hitters have pumped billions and hired some of the greatest mathematical minds to develop models that outperform the markets and ... they have not succeeded."
In fact as you can read here:
http://en.wikipedia.org/wiki/Algorithmic_trading

"A third of all EU and US stock trades in 2006 were driven by automatic programs, or algorithms, according to Boston-based consulting firm Aite Group LLC. By 2010, that figure will reach 50 percent, according to Aite.[2]

In 2006 at the London Stock Exchange, over 40% of all orders were entered by algo traders, with 60% predicted for 2007. American markets and equity markets generally have a higher proportion of algo trades than other markets, and estimates for 2008 range as high as an 80% proportion in some markets. Foreign exchange markets also have active algo trading (about 25% of orders in 2006).[3] Futures and options markets are considered to be fairly easily integrated into algorithmic trading[4], with about 20% of options volume expected to be computer generated by 2010.[5] Bond markets are moving toward more access to algorithmic traders.[6]"

So not only this field is alive, it is also rapidly increasing. And if it is increasing maybe there could be something valuable in this approach.
Well, now I don't know if automatic trading requires more intelligence or Nobel prizes or money or resources in general than discretionary trading.
What I know for sure is only these:
1) At my age you know what are your strengths and weaknesses and I know for sure that I'm absolutely not more intelligent than the average person on this planet.
2) In years of testing, trying + live, real trading I've never been able to master any discretionary system such in a way to be profitable for more than 2 consecutive weeks.
3) But at the same time I've developed 10 automatic systems that show consistent profitability with small drawdowns over 8 years of training data, 6 months of out of sample data and (for three of them, the only three that I decided to trade live) 7 months of profitable live trading.
You know, all of this does really prove anything at all. Too few data and too short time to prove anything. But, please, tell me sincerely, if you was in me, given those facts, would you go for discretionary trading or automatic?
These dualistic arguments rarely are that black and white and have a tendency to become polarized. I suspect that the marriage of the technical precision of the machine and the judgment of the human brain complement each other way beyond what either individual method could achieve.

I realise my folly now, perhaps I should have said that these mathematical models had failed to outperform what a good trader can produce. Even so, I suspect that this may be incorrect too; although it is harder than one might think to prove this conclusively. The machines can work all day and night and don't take breaks or holidays - and are ultimately less expensive than human traders so it is no surprise that they have been adopted by fund managers. The idea of developing an EA that could provide a consistent passive income stream is very attractive and I can see why people spend a lot of time attempting to do exactly that. However my experience is that the best systems are not entirely programmable, the car needs a driver to use his judgment to steer it in the right direction.

In answer to your question, I guess it is all about being aware of your strengths and weaknesses. If you are profitable in one area and not in another then it makes sense to exploit that area. I absolutely would use the EA route were I in your shoes.

I very much doubt you are below average intelligence though. The average person is pretty stupid - and to think that 50% of people are below average intelligence!
Quoting birdt
However my experience is that the best systems are not entirely programmable, the car needs a driver to use his judgment to steer it in the right direction.
You know, it all depends on what your trying to achieve.
For example there are also systems in other fields where humans simply cannot do anything at all without computers. There are for example some military airplanes that in order to achieve special performances are designed in such a way that no human pilot can fly them without the constant and realtime assistance of some computers. Should the onboard computers stop working those airplanes would become so unstable that they would crash and burn in a matter of half a second, despite how good the pilot is.
Human brain is surely the most intelligent tool we know so far. But this does not mean that it is always the best tool one could use to solve any kind of problem. Sometime using it in an indirect way (usign it to develop other intelligent tools) could prove to be a better approach.
It has happened for example with neural network at least when applied to insurance companies and the like. It is a fact that no insurance company now a day would work without using some sort of neural network to better drive their business. And the profit increase is around 10-12% in those fields.
And of course there are other fields in which computers are definitively worse than humans. For example I've never heard any good song completely composed by a computer.
.Ubeee.
Wow I didn't thin I was going to start this kind of discussion, but glad it turned into it. What I am trying to do is revise my trading a little. And yes I am not one to give up easily. It became apparent that the EA I was using worked fairly well in a ranging market, When the USD decided it wanted to
do it's thing, it kind of sent it into a spin, I just didn't catch it in time. And on the manual trading portion of my trading, yes I have looked at it and I am still dissecting the loosing trades. Many of which I should have taken S&R into account, and not try to go long just under resistance, or go short just over support. Something I should have incorporated while ago.

Rusty
  • Post #50
  • Quote
  • Last Post: Sep 1, 2008 10:36am Sep 1, 2008 10:36am
Quoting rusty105
Ok, i will post this in the rookie section, cause it was rookie thinking that got me to this point. I was doing good, was profitable for a few months, had a few good runs, and thought I coould do this easily, then it hit. I use a combo of an EA on a few pairs, with what I thought was low risk, to gather pocket change all day, while I wait for the better signals from a different method (DIBS). and this was working pretty well. I even took $$ out of my trading account to use for something else That is about the time it happened, Several pairs hit a strong trend, and the EA got caught, while I was still doing pretty good with DIBS (>500 pip on GBPUSD) it couldn't out weight the losses in the EA's. So now I am down 80% or so (small account only couple hundred US) What I think got me here was exposing my account to too many pairs, and too many open trades at once. And not paying attention to my manual trades, thinking I could just blindly open an order, and it would work out. I am not worried about the $$ like I said it was only a few hundred (I would spend more on a weekend skiing) But the roller coaster emotions are tough, one day i am up 20%, next day down 40%, then down 80% total. What does everyone tell themselves when they hot a loss like that ?? Do you take a few days/weeks ff, or just keep on trading? I WILL NOT give up, but I wont go broke trying either.

What does everyone do when they hit the "Wall"??

Rusty
http://www.forexfactory.com/showthread.php?t=103840
Trading is 80% Emotional, 20% Technical
0 traders viewing now